LibreTexts JupyterHub Default Environment Lite

This repository contains the repo2docker config files for the lite Jupyter image — a Python-only stack optimized for fast cold-starts. It is used by the CKEditor-Binder-Plugin to power thebe cells embedded in libretexts.org pages.

For the full multi-language stack (Python, R, Octave, Sage, Julia) used at jupyter.libretexts.org, see LibreTexts/default-env.

Because the config files live at the repo root, this repo can be handed directly to repo2docker with no --subdir:

jupyter-repo2docker .

The main branch is not used directly to source production images. Build manually with ./build.sh, push to the docker registry, and then update the relevant helm values to roll out. CKEditor-Binder-Plugin directly references git tags in this repo.

Repo2docker Configuration Files

• environment.yml — conda packages. We only install from the conda-forge channel.
• apt.txt — Debian packages via apt-get. Restricted to Ubuntu 18.04 (bionic) repositories.
• postBuild — shell script run after everything else is installed; pre-warms import caches.

You can read more about each configuration file's purpose from the repo2docker documentation.

Building locally

./build.sh

To reuse layers from a previous build (faster rebuilds):

docker pull libretextsbinder/default-env-lite:<prev-hash>
PREV_TAG=<prev-hash> ./build.sh

Keeping in sync with the full image

The scipy-stack pins (numpy, scipy, matplotlib, pandas, sympy, cython) match the full image so notebooks behave identically across both. When you bump a pin here, bump it in the full image too — and vice versa.

Note that the lite image tracks a current Python (3.11), while the full image is held back to 3.8 by sage. The shared scientific-stack pins are chosen to resolve on both.